Improvement of dead time and decoding resolution for position-sensitive detectors using a fully dynamic approach of light collection

Abstract

Photo-sensor sharing (multiple blocks or crystals) can achieve high-resolution position-sensitive detectors but it also increases the dead time and pileups for scintillation event detection. Several methods such as pulse-clipping and HYPER (high-yield-pileup-event-recovery) have been introduced to minimize the dead time and pileups with a trade-off of less scintillation light collection. However, collecting smaller number photoelectrons would increase the statistical error, which in a turn will decrease the decoding resolution. In this study, instead of applying the HYPER method to 3 Anger-signal (X, Y and E) simultaneously, we use an individual dynamic approach for each photo-sensor to maximize the scintillation light collection while it still has a capability of rejecting pileups. The photo-electron collection for one photo-sensor involving a current event decoding will not be disturbed until a new event is detected that also requires this photo-sensor for position-decoding. If a new event comes from an adjacent detector block only sharing one photo-sensor currently involving the previous event decoding, it may only disturb the light collection of this sharing photo-senor; hence only one photo-sensor creates a poor statistical error and the rest photo-sensors can still collect a large number of photo electrons with good statistics for the previous event to achieve a good decoding resolution. This paper compares the decoding results using pulse clipping and this new proposed fully dynamic approach at various count-rates for a regular position-sensitive block detector and a PMT-quadrant-sharing (PQS) block detector. To study the decoding resolution at various high count-rates, a pulse waveform library was built by recording a large number of pulses by a digital oscilloscope from a detector test-bench at a low count-rate first and then boost to different high count-rates by software generated Poisson event time sequence. The result shows this fully dynamic approach reduces the dead space of PQS detectors by x2.25 without increasing the statistical noise.